畳み込みニューラルネットワークを用いた熱源システムの自動不具合検知・診断の実装と学習済みネットワークの類似システムへの適用検討

抄録

The installation of Automated Fault Detection and Diagnosis (AFDD) using Convolutional Neural Network (CNN) was first introduced. Next, case studies were investigated to deploy this AFDD method to similar systems. The results showed the difficulty to use a trained CNN for a specific system directly to other systems, however, the transfer learning was effective to reduce computational costs.